Activated Carbon Derived from Spherical Hydrochar Functionalized with Triethylenetetramine: Synthesis, Characterizations, and Adsorption Application

Green Process Synth 2017; 6: 565–576

Hai Nguyen Tran, Fu-Chuang Huang, Chung-Kung Lee and Huan-Ping Chao* Activated carbon derived from spherical hydrochar functionalized with triethylenetetramine: synthesis, characterizations, and adsorption application

DOI 10.1515/gps-2016-0178 Keywords: activated carbon; dye; glucose; heavy metal; Received October 15, 2016; accepted January 2, 2017; previously hydrochar; triethylenetetramine. published online March 6, 2017

Abstract: This study investigated the adsorption capacities of various contaminants on glucose-derived hydrochar (GH) and glucose-activated carbon (GAC) 1 Introduction functionalized with triethylenetetramine (TETA). The Activated carbon (AC), with its exceptionally large specific two-stage synthesis process consisted of (1) hydrother- surface area, high pore volume, well-developed inter- mal carbonization using various TETA concentrations nal porous structure, and abundant surface functional (1%–5%) to create TETA-functionalized GHs, and (2) groups (polar characters), has been widely applied in chemical activation with NaOH to produce TETA-GACs. various industrial processes. In water treatment, ACs are The basic properties of the adsorbents were examined considered effective adsorbents for the removal of various using Brunauer-Emmett-Teller (BET) surface area analy- organic contaminants. According to an industry market sis, Fourier transform infrared (FTIR) spectrometry, scan- research report [1], the global demand for AC is estimated ning electron microscopy (SEM), and energy dispersive to increase 8.1% per year, and be up to 2.1 million metric X-ray (EDX) spectroscopy. The adsorption characteristics tons by 2018. Nevertheless, the high cost of commercial of the GH and GAC samples toward two heavy metal ions ACs restricts their large-scale use in industries. (Pb2 + and Cu2 +), phenol, methylene green (MG5), and acid Notably, the morphology of AC plays a key role in red 1 (AR1) were also examined. The results indicated that its application; various conformations comprise pow- GAC and GH exhibited excellent adsorption capaci- 1% 1% dered and granular AC, AC fibers, carbon monoliths, ties. Specifically, the maximum adsorption capacities of carbon hollow spheres, carbon nanotubes, and, carbon GAC and GH reached 370 mg/g and 128 mg/g for Pb2 +, 1% 1% spheres [2]. Spherical carbon can be obtained through 208 mg/g and 84 mg/g for Cu2 +, 196 mg/g and 137 mg/g hydrothermal carbonization of diverse organic materials for phenol, 175 mg/g and 67 mg/g for MG5, and 156 mg/g (polyvinylpyrrolidone, sucrose, xylose, fructose, furfural, and 21 mg/g for AR1, respectively. In conclusion, amine glucose, starch, saccharose, and cellulose) in a controlled functionalization on the surface of GHs and GACs effi- temperature autoclave (150–350°C) for 2–48 h at a specific ciently enhances the removal capacities of various con- pressure for producing hydrochar spherical microparti- taminants in water. cles [3–6]. Hydrochars have been commonly tailored for manufacturing ACs with desired characteristics because of the unique attributions of hydrochar – namely a high *Corresponding author: Huan-Ping Chao, Department of density of oxygenated functional groups and a low Environmental Engineering, Chung Yuan Christian University, degree of condensation and impurity [3]. Compared with Chungli 32023, Taiwan, e-mail: [email protected] Hai Nguyen Tran: Department of Civil Engineering, Chung Yuan other ACs, spherical ACs exhibit several enhanced char- Christian University, Chungli 32023, Taiwan; and Department acteristics such as high wear resistance, high mechani- of Environmental Engineering, Chung Yuan Christian University, cal strength, superior adsorption, high purity, low ash Chungli 32023, Taiwan content, smooth surface, low pressure drop, high bulk Fu-Chuang Huang: Department of Environmental Technology and density, high micropore volume, and controllable pore Management, Taoyuan Innovation Institute of Technology, Chung-Li, 32091, Taiwan size distribution [2, 3]. Chung-Kung Lee: Department of Environmental Engineering, Vanung D-glucose (a simple carbohydrate) is the most abun- University, Chung-Li, 32061, Taiwan dant sugar unit in biomass, and it is the major product 566 H. N. Tran et al.: Activated carbon derived from spherical hydrochar functionalized with triethylenetetramine of lignocellulosic biomass acid hydrolysis. There- synthesized through a NaOH chemical activation of the fore, D-glucose is the most used precursor to produce GH samples; GH and GAC samples without TETA modifi- hydrothermal carbonization [4]. A review of the extant cation were also simultaneously prepared. In addition, literature revealed that the adsorption capacity of glu- the basic properties of the adsorbents were examined cose-derived carbon spheres is enhanced by modifying using several techniques (i.e. Brunauer-Emmett-Teller their surfaces with various surfactants. For example, [BET] surface area, Fourier transform infrared [FTIR], Demir-Cakan and others [7] prepared carbonaceous and scanning electron microcopy-energy dispersive materials (hydrochar) through the hydrothermal car- X-ray [SEM-EDX] analyses). Furthermore, the adsorption bonization of glucose in the presence of acrylic acid, behaviors of GHs and GACs with and without a TETA- and they concluded that carboxylate-rich absorbents functionalized surface for two heavy metals (Pb2 + and were successfully employed for the removal of Cd2 + and Cu2 +), phenol, a basic dye (methylene green 5 [MG5]), Pb2 + from water. Wang and coworkers [8] reported on a and an acid dye (acid red 1 [AR1]) were conducted in one-step hydrothermal preparation of amino-function- batch experiments. alized carbon spheres (hydrochar) by mixing a glucose– ammonia solution at low temperature to improve their adsorption performance toward Cr(VI). Various nitrogen sources, derived from nitrogen gas, 2 Materials and methods ammonia, ammonia gas, amines, urea, pyridine, ace- 2.1 Preparation of adsorbents tonitrile, melamine, dimethylformamide, 2-amino-4,6-di- chloro-s-triazine, benzylamine, triethylenetetramine The synthesis processes of carbonaceous adsorbents with and with- (TETA), ethylenediamine, and polyazomethineamide, out surface modification are illustrated in Figure 1. First, a mixture of have been used for synthesizing nitrogen-doped porous glucose and TETA powders (purchased from Merck) was completely carbons [8–13]. Of these, TETA has been widely applied as dissolved in 150 ml of distilled water and then transferred into a 200- an effective cationic surfactant to modify an adsorbent’s ml Teflon-lined autoclave. Mass ratios of glucose and TETA ranging surface; for example the effective adsorption of Cu2 +, Cd2 +, from 1% to 5% were used for the modification. After a 48-h hydrothermal process at 190°C, the remaining brown precipitate (hydrochar) and Pb2 + from aqueous solutions by both succinylated particles were separated using vacuum filtration, washed repeatedly twice-mercerized sugarcane bagasse and succinylated with a 95% alcohol solution, and then washed in the distilled water mercerized cellulose modified with TETA has been pro- until the pH of the filtrate reached approximately 7.0. The hydrochar posed by [12, 13]. Karnitz and colleagues [10] also com- samples were then collected and dried in an oven at 105°C for 24 h. pared the adsorption capacities of Cu2 +, Cd2 +, and Pb2 + For convenience, the GH samples modified with TETA were labeled from aqueous systems by sugarcane bagasse chemically modified with the polyamines ethylenediamine and TETA.

In addition, Barsanescu and colleagues [14] investigated Glucose the effectiveness of using acrylic copolymer prepared Glucose (%)/TETA (%) = from organic matrices with distinct crosslinking degrees, (95/5; 96/4; 97/3; 98/2; 99/1) followed by ethylenediamine and TETA functionalization 2 + Triethylentetramine on the surface to remove Zn ions. However, the char- Water (TETA) acterizations of TETA-modified spherical and activated carbons derived from glucose and the efficiency of these Hydrothermal adsorbents toward the removal of organic and inorganic (190°C, 48 h) contaminants have not been examined and reported Hydrochar Hydrochar (GH , GH , elsewhere. (GH ) 1% 2% 0% GH , GH , GH ) The current study synthesized a TETA-modified 3% 4% 5% adsorbent to determine the adsorption of inorganic and Pyrolysis 5% NaOH organic compounds from aqueous solutions. Glucose- (800°C, 3 h) 5% NaOH derived hydrochar (GH) samples, functionalized with Activated carbon various TETA concentrations (1%–5%), were prepared Activated carbon (GAC1%, GAC2%, GAC3%, (GAC0%) through hydrothermal carbonization of the glucose- GAC4%, GAC5%) TETA mixture, and they were subsequently used as the precursors to produce glucose-AC (GACs). The GAC Figure 1: Schematic illustration of the preparation procedure for the samples with nitrogen groups on their surface were adsorbents. H. N. Tran et al.: Activated carbon derived from spherical hydrochar functionalized with triethylenetetramine 567

GH , GH , GH , GH , and GH , according to the ratio of TETA to 5% 4% 3% 2% 1% 3 Results and discussion glucose. A GH sample without TETA (GH0%) was synthesized under the same conditions. The AC was prepared by first impregnating approximately 10 g of the hydrochar sample with 100 ml of NaOH solution (5%). Next, 3.1 Characterization of adsorbents the mixture of hydrochar and NaOH was dried in an oven at 105°C until the solution completely evaporated. The pyrolysis process was 3.1.1 Textural properties conducted in a column-type stainless steel tube in an inert atmos- phere; subsequently, the tube was heated to 800°C at a heating rate Figure S2 depicts the variation of N2 adsorption-desorption of 5°C/min and the temperature was maintained for 3 h under a N 2 isotherms for GACs synthesized with various glucose-TETA flow rate of 100 ml/min. The GACs were washed with alcohol and the ratios, clearly indicating that the adsorption isotherms of distilled water until the pH of the filtrate was approximately 7.0, and then they were dried in an oven at 105°C for 24 h. The final GAC sam- all the GACs generated hysteresis loops. Isotherms are ples were labeled according to their precursors (i.e. GAC5%, GAC4%, typically characterized by mesopores and macropores,

GAC3%, GAC2%, GAC1%, and GAC0%). with a small external surface area [15]; moreover, a wide- knee hysteresis loop was present in both the adsorption and desorption isotherms, appearing in the multilayer 2.2 Adsorbent characterization range of physical adsorption isotherms. According to the nomenclature of the International Union of Pure and The textural properties of the GACs were obtained from a conven- Applied Chemistry, porous carbon materials exhibit an tional analysis of nitrogen adsorption–desorption isotherms, which H4-type hysteresis loop, which is associated with narrow, were measured at 77 K by a sorptometer (Micromeritics ASAP 2020). FTIR spectroscopy (FT/IR-6600 Jasco) was then performed to detect slit-like pores. In addition, the AC sample without TETA the functional groups present on the adsorbent surface, and the (GAC0%) exhibits a relatively higher non-micropore volume adsorbent particles mixed with KBr were subsequently pelleted. (0.167 cm3/g) than the ACs samples modified with TETA Finally, the morphology and element composition of the adsorbents 3 (GAC1%−5%; 0.033–0.108 cm /g). The reason is ascribed to were examined using SEM (S-3000N, Hitachi) at 10 kV and EDX spec- TETA fills in the pores. The addition of TETA can alter the troscopy (EDX), respectively. surface properties of GAC1%−5%. Table 1 provides a summary of the corresponding tex- 2.3 Adsorptive isotherm experiment tural parameters of the GACs. Increasing the TETA ratios resulted in a decreased surface area (SBET), non-micropore Two metal ions (Pb2 + and Cu2 +) and three organic compounds (phe- volume, and total pore volume, but an increase of the nol, AR1, and MG5) were selected as typical adsorbates to determine micropore volume. Furthermore, the average pore width the adsorption characteristics of the GH and GAC samples. The chem- decreased as the ratio of TETA increased, indicating that ical structures and basic properties of phenol, MG5, and AR1 are rep- the GAC pore size was determined by the amount of TETA 2 + resented in Figure S1, and the basic and ionic properties of Cu and in the sample. As expected, the average pore size of GACs Pb2 + are summarized in Table S1. was a function of the proportion of the non-micropore To avoid the precipitation of ionic salts and effect of dissociation on the selected organic solutes at high pH values, the initial pH val- (mesopore and macropore) volume. ues of the solutions were adjusted to 5.0 ± 0.1. Approximately 0.2 g of the adsorbent was subsequently added to 50 ml of aqueous adsorbate solution in a Teflon centrifuge tube. To prevent adsorption competition, only a single adsorbate was added in the centrifuge tube. 3.1.2 Morphological properties The centrifuge tubes were then placed in a reciprocating shaker with 180 rpm and equilibrated for 48 h at 30°C. After the completion of The morphologies of GH0% and GAC0% are portrayed in the adsorption process, each solution was centrifuged for 30 min and Figure 2. The SEM images of both GH and GAC reveal strained through a 0.2-μm filter. The concentrations of phenol, MG5, 0% 0% interconnected spheres with relatively uniform sizes, and AR1 were determined using ultraviolent-visible spectrophotom- etry (Genesys 10 UV-Vis; Thermo Scientific) at maximum wavelengths smooth outer surfaces, and regular spherical shapes. of 265 nm, 655 nm, and 530 nm, respectively. Atomic absorption spec- According to a study by Sevilla and Fuertes [17], the for- trometry (Avanta, GBC) was used to determine the concentration of mation of a carbon-rich solid through the hydrothermal 2 + 2 + Pb and Cu ions. carbonization of glucose is attributed to dehydration, Each experiment was performed in duplicate, and blank sam- condensation, or polymerization and aromatization reac- ples without the adsorbent were also conducted simultaneously. If the bias of the repeated experiment exceeded 15%, a triplicate run tions. In short, the carbon-rich hydrochar particle consists was performed. All chemicals used in this study were of analytical of two parts: a hydrophobic core comprising a highly aro- reagent grade. matic nucleus, and a hydrophilic shell comprising a high 568 H. N. Tran et al.: Activated carbon derived from spherical hydrochar functionalized with triethylenetetramine

Table 1: Textural parameters of the glucose activated carbon (ACs) and commercial AC.

a GAC5% GAC4% GAC3% GAC2% GAC1% GAC0% CAC BET surface area (m2/g) 260 313 256 288 233 335 768 Langmuir surface area (m2/g) 311 402 327 364 295 418 – External surface area (m2/g) 69 113 115 135 151 195 – Micropore surface area (m2/g) 191 200 141 153 82 140 – Total pore volume (cm3/g) 0.110 0.144 0.149 0.169 0.137 0.217 0.43 Micropore volume (cm3/g) 0.077 0.088 0.058 0.061 0.029 0.050 0.21 Non-micropore volume (cm3/g) 0.033 0.056 0.091 0.108 0.108 0.167 0.22 Micropore/total pore volume (%) 70.27 60.88 38.78 36.25 20.98 22.98 48.84 Average pore width (nm) 1.692 1.843 2.332 2.343 2.354 2.588 1.63

Total pore volume = sum of micropore volume and non-micropore volume. aThe experimental data were published in [16]. BET, Brunauer-Emmett-Teller; CAC, commercial activated carbon ; GAC, glucose-activated carbon.

As displayed in Figure 3, there were substantial changes to the surface morphology of the samples following TETA modification. The samples initially possessed extremely rough surfaces, and the results demonstrated that TETA was successfully grafted onto the surface of GAC during our experiment. Moreover, the morphologies

of GH1% and GAC1% were nearly identical, again confirm- ing that chemical activation with NaOH did not cause noticeable alterations to the morphology of GACs. There- fore, NaOH activation only modified the surface of GACs, without affecting the morphology and the bulk carbon core. This conclusion is consistent with findings in the extant literature [20].

Figure 2: Scanning electron microcopy-energy dispersive X-ray

(SEM-EDX) spectra of the (A) glucose-derived hydrochar (GH)0% and

(B) glucose-activated carbon (GAC)0% samples. concentration of reactive oxygen functional groups such as hydroxyl-phenolic, carbonyl, or carboxyl [17, 18].

Figure 2 also indicates that GH0% exhibited a higher average diameter (0.82–0.92 μm) than that of GAC0% (0.38– 0.43 μm), suggesting that the size of carbon spheres is narrowly affected by the chemical activation process; the SEM results also revealed that the spherical morphology of

GAC0% possesses a similar particle size. This means that the regular shape of carbon spheres was still maintained after chemical activation. Interconnected particle properties of

GH0% and GAC0% afforded more facile adsorbent separation Figure 3: Scanning electron microcopy-energy dispersive X-ray from aqueous solutions (i.e. filtration with a microfilter) (SEM-EDX) spectra of the (A) glucose-derived hydrochar (GH)1% and because of the overall increase in particle size [19]. (B) glucose-activated carbon (GAC)1% samples. H. N. Tran et al.: Activated carbon derived from spherical hydrochar functionalized with triethylenetetramine 569

The EDX spectra displayed in Figures 2 and 3 reveal in hydrochar in the form of –OH. The presence of aro- that elements containing C and O are primary elements; matic C–H out-of-plane bending vibrations is evident for example, hydochar is a carbon-rich solid that is by the peak at 798 cm − 1 (including aldol condensation approximately 93–96% carbon. The higher percentage and dehydration) [7]. As researchers have previously of Na in the GAC samples indicated that the presence of noted, the appearance of C=O groups on the surface of neither the functional hydrolyzed-carboxylic acid group glucose hydrochar is due to the dehydration of equatorial (–COO–Na + or –COO − under aqueous conditions) nor the hydroxyl groups [3, 17]. + − functional hydrolyzed-phenolic group (–O–Na or –O The FTIR spectrum of GAC0% (Figure S3) revealed under aqueous conditions) was strongly correlated with similar peaks to those of GH0% (i.e. C=O, C=C, C–O, –OH, the high cationic affinity of the GAC samples [19]. The and C–H). New vibration bands at 2919 cm − 1 and 2850 cm − 1 presence of carboxylic and phenolic groups on the surface corresponded to asymmetrical and symmetrical C–H, of GHs and GACs was additionally confirmed by the FTIR respectively. The peaks related to the carboxyl groups in analysis results (Figure 4). Moreover, the higher percent- GAC0% noticeably shifted toward lower wavelengths after age of oxygen in the GAC samples suggests that the GAC the sample underwent chemical activation with NaOH, samples had more acidic functional groups than the GH compared with the peaks in GH0% groups; specifically, the samples. Noticeably, the percentage of N element is lower C=O and C–O peaks decreased from 1659 cm − 1 to 1590 cm − 1 than the detection limitation because of low impregnation and from 1200 cm − 1 to 1030 cm − 1, respectively. In addi- of TETA (only 1%); therefore it cannot be identified in EDX tion, the marked increase in band intensity at 3390 cm − 1 analysis. indicated that new –OH groups can form during chemical activation. Qualitative information on the functional groups

3.1.3 Functional group characterizations present on the TETA-modified GACx% surfaces and their spectroscopic assignments is displayed in Figure 4. The − 1 The functional groups of GAC0% and its precursor bands observed at approximately 3370 cm are attrib-

(GH0%) were identified using FTIR (Figure S3). For the uted to the stretching vibrations of the primary amine − 1 GH0% sample, two peaks identified at 1965 cm and (–NH2) group overlapped with the stretching band in the 1200 cm − 1 were involved in the stretching of the C=O hydroxyl groups (O–H) centered at 3400–3500 cm − 1 [21]. and C–O bonds in “pristine” carboxyl groups, respec- The bands found in the 2800–3000 cm − 1 range belong to tively (Figure S3). Furthermore, the band at 1600 cm − 1 is asymmetrical and symmetrical C–H stretching vibrations related to the aromatic C=C ring stretching motion, and of the methyl (–CH3–) and methylene (–CH2–) groups. The the band at 3400 cm − 1 confirms the presence of oxygen presence of a carbon-carbon triple bone (C≡C) in disub- stituted alkynes can also be inferred from the bands in the range of 2250–2400 cm − 1. The sharp peaks recognized between 1500 cm − 1 and 1690 cm − 1 are attributed to the –NH –CHn- –C≡C– C=O C=C C−O C−H 2 stretching vibration of C=O in the secondary amide group [22] or –NH scissor frequencies [21, 23]. The C=O bond in GAC5% 2 the carboxyl group clearly shifted to lower wavelengths 1583 3370 1037 (i.e. deformation vibrations) after reaction with TETA. GAC4% 1406 Specifically, the C=O bond in the pristine carboxyl group 1586 3374 1035 peaked at 1590 cm − 1 (Figure S3), but after TETA modifi- GAC3% 1403 cation, the peak decreased to approximately 1580 cm − 1 1582 GAC 3379 1031 − 1 Transmittance (%) 2% 1414 (Figure 4). The bands observed at around 1400 cm also

1583 reveal the aromatization (C=C) of the GACx% samples. GAC 3376 1033 − 1 1% 1400 Finally, the bands located between 960 cm and − 1 1583 1130 cm are characteristic of stretching C–O groups; 3369 1405 1022 the low intensity in these bands confirms the existence 4000 3500 3000 2500 2000 1500 1000 650 of strong reactions between pristine carboxyl groups and Wavenumber (cm – 1) TETA. The reaction between the carboxyl groups and = Figure 4: Fourier transform infrared (FTIR) spectra of the glucose TETA, which converts the C O in carboxyl groups into activated carbon (AC) samples at various triethylenetetramine (TETA) C=O in the amide groups, was introduced by Brady and impregnation ratios. Duncan [24] and occurs as follows: 570 H. N. Tran et al.: Activated carbon derived from spherical hydrochar functionalized with triethylenetetramine

O H O Carbodiimide —C — N — CH2CH2(NHCH2CH2)2NH2 (1) – C — OH + C6H18N4 + H O Carboxylic (TETA) Amide 2

3.2 Adsorption isotherms 3.2.1 Adsorption isotherms for lead and copper

In this study, the Langmuir model [Eq. (2)] and Freundlich Figures 5 and 6 show the adsorption isotherms of Pb2 + and model [Eq. (3)] were applied to describe the adsorp- Cu2 + ions on the hydrochars and ACs at various impreg- tive behavior of the contaminants on the synthesized nation ratios of TETA and glucose. After the surfaces of o adsorbents: the GH samples were modified with TETA, the Q max values 2 + 2 + 0 of Pb and Cu increased by 29%–114% and 15%–245%, QKmaxLCe qe = (2) respectively. This result indicates that the TETA surfactant 1 + KC Le effectively bonds metal ions. The maximum adsorptive 2 + amounts of the Pb ions were ordered as follows: GH1% > G 1/n H3% > GH2% > GH4% > GH5% > GH0%. Conversely, the ordering qKeF= Ce (3) 2 + of Cu adsorption was GH 5% > GH1% ≈ GH3% > GH2% > GH0%. where qe (mg/g) is the amount of adsorbate uptake at equilibrium (which was calculated from the mass balance equation), C (mg/l) is the adsorbate concentra- e A 260 o GH 5% tion at equilibrium, Q max (mg/g) is the maximum satu- 240 GH 4% 220 rated monolayer adsorption capacity of the adsorbent, GH 3% 200 KL (l/mg) is the Langmuir constant related to the affinity GH 2% 1/n 180 GH 1% between an adsorbent and adsorbate, KF (mg/g)/(mg/l) 160 GH 0% is the Freundlich constant (which characterizes adsorp- Langmuir fit 140 tion strength), and 1/n (dimensionless; 0 < n < 10) is a

(mg/g) 120 e q Freundlich intensity parameter (which indicates the 100 magnitude of the adsorption driving force or surface 80 heterogeneity). 60 The coefficient of determination (R2) of the nonlinear 40 optimization method was computed using the following 20 equation: 0 0 100 200 300 400 500 600 700 2 ()qq− C (mg/l) ∑ e, cal e R2 =−1 e, exp − 2 B 260 ∑()qqe, expe, mean 240 2 ()qq− 220 = ∑ e, cale, mean (4) −+22− 200 ∑∑()qqe, cale, mean ()qqe, cale, exp 180 160 where qe, exp (mg/g) is the amount of adsorbate adsorbed 140

(mg/g) 120 at equilibrium obtained from the experiment; qe, cal (mg/g) e GAC 5% q is the amount of adsorbate uptake determined from the 100 GAC 4% 80 GAC 3% model after using the Solver add-in, and qe, mean (mg/g) is 60 GAC 2% the mean value of qe, exp [25]. GAC 1% 40 The Langmuir and Freundlich parameters are GAC 0% 20 described in Tables S2–S6. Notably, the determination Langmuir fit 0 coefficients (R2) of the Langmuir model were higher than 0 100200 300400 500600 700 those of the Freundlich in most of the cases, indicating C (mg/l) that the adsorption characteristics of the contaminants e in the GH and GAC samples were adequately described by Figure 5: Adsorption isotherms of Pb2 + ions by various (A) glucose the Langmuir model. hydrochar and (B) activated carbon (AC) samples. H. N. Tran et al.: Activated carbon derived from spherical hydrochar functionalized with triethylenetetramine 571

A 280 activation with NaOH can effectively increase the cation 260 GH 5% GH 4% exchange capacity. Furthermore, other adsorption mecha- 240 GH 3% nisms – complex reactions with oxygen functional groups 220 GH 2% 200 GH 1% (–COOH and –OH) and pore filling – might improve the GH 0% 180 Langmuir fit binding capacities for GACx% samples. 160 o The Q max values of GACs modified with TETA are 140

(mg/g) slightly higher than those without TETA, appropriately e

q 120 2 + 2 + 100 6%–31% for Pb and 2%–20% for Cu . Therefore, we 80 can conclude that neither amine groups had minor con- 60 tributions in the adsorption mechanisms of the surface 40 modified GACs samples nor several amine groups might 20 be destroyed during the carbonization at 800°C. The 0 maximum diminished theoretical adsorption capacities 0 100 200 300 400 500 600 700 800 900 1000 for Pb2 + were ordered as follows: GAC > GAC > GAC C (mg/l) 3% 1% 4% e > GAC > GAC > GAC ; by contrast, the ordering of Cu2 + 280 5% 2% 0% B > = > > > 260 was GAC5% GAC4% GAC2% GAC1% GAC3% GAC0%. o 240 Generally, the adsorption efficiencies (Q max; mmol/g) 220 of Cu2 + were remarkably higher than those of Pb2 +, in both 200 GHx% and GACx% (Table 2). The adsorption amount order 180 can be explained by the chemical properties of the ions 160 140 GAC 5% (i.e. hydrated ionic radius, ionic potential, electronegativ- (mg/g) e GAC 4% ity, charge density, first hydrolysis equilibrium constant, q 120 100 GAC 3% and ionic radius) [28], which are summarized in Table S1. GAC 2% 80 GAC 1% As expected, most of the GH and GAC samples that 60 GAC 0% were functionalized with TETA in this study provided 40 Langmuir fit 2 + 2 + 20 greater adsorption capacities for Cu and Pb compared 0 with commercial AC (Table 2) under the same experimen-

0 50 100150 200250 300 tal conditions, indicating that TETA is efficient in binding metal cations in the solutions. Ce (mg/l)

Figure 6: Adsorption isotherms of Cu2 + ions by various (A) glucose hydrochar and (B) activated carbon (AC) samples. 3.2.2 Adsorption isotherms for phenol, MG5, and AR1

The adsorption isotherms of phenol, MG5, and AR1 on the

If the oxygen-containing groups on the surface of GH0% GH and GAC samples are depicted in Figures 7–9. For the

(–COOH and –OH) are assumed to take primary responsi- samples modified with TETA (GH1%−5% and GAC1%−5%), their bility for binding Cu2 + and Pb2 + ions, the enhanced adsorp- adsorption capacities toward phenol, MG5, and AR1 were tion capacities of the metal cations for GH1%−5% are mainly inversely proportional to the glucose-TETA mass ratios attributable to the complex reactions between Cu2 + and (Table 2). The adsorption capacities of TETA-modified 2 + Pb ions with the amino groups (–NH2); other scholars GACs decreased in the following order: GAC1% > GAC2% > have confirmed this finding [26, 27]. As expected, the GH GAC4% > GAC3% > GAC5% for phenol, GAC1% > GAC2% > GAC4% > 2 + samples exhibited higher adsorption capacities of Pb GAC5% > GAC3% for AR1, and GAC1% > GAC2% > GAC3% > GA 2 + and Cu than the commercial AC did (Table 2). C4% > GAC5% for MG5. A similar decreasing tendency was

The GAC1%−5% samples with TETA-modified surfaces found in the TETA-modified GH samples. possessed higher adsorption capacities toward heavy Several possible mechanisms of the phenol com- metals compared with their precursors (GH1%−5%), with pound adsorption have been postulated in the literature increasing ratios of approximately 189–316% for Pb2 + [29–31]: (1) electrostatic attractions, (2) π–π dispersion and 160–720% for Cu2 + (Table 2). Notably, the adsorption interactions, (3) hydrogen bonding, and (4) pore filling. 2 + 2 + capacity of GAC0% for Pb and Cu was greatly enhanced, The mechanisms of electrostatic attractions between + compared with that of GH0%, increasing to 391% and the positive protonated amine groups (–NH3 ) on the − 704%, respectively. The result demonstrates that chemical surface of adsorbents and phenolate anions (C6H5O ) are 572 H. N. Tran et al.: Activated carbon derived from spherical hydrochar functionalized with triethylenetetramine

o Table 2: Maximum monolayer adsorption capacities (Q max; mg/g) of the glucose-derived hydrochar (GH), glucose-activated carbon (GAC) and CAC samples toward Pb2 +, Cu2 +, phenol, methylene green 5 (MG5), and acid red 1 (AR1).

Adsorbents Pb(II) Cu(II) Phenol MG5 AR1

GH5% 77.5 (0.374) 89.3 (1.405) 38.2 (0.406) 27.5 (0.064) 17.0 (0.033)

GH4% 93.5 (0.451) 29.8 (0.468) 63.3 (0.673) 55.9 (0.129) 30.2 (0.059)

GH3% 115 (0.555) 82.0 (1.290) 74.6 (0.793) 53.5 (0.124) 17.6 (0.035)

GH2% 90.9 (0.439) 35.7 (0.562) 104 (1.107) 58.8 (0.136) 33.1 (0.065)

GH1% 128 (0.619) 84.7 (1.334) 137 (1.456) 67.6 (0.156) 36.1 (0.071)

GH0% 59.9 (0.289) 25.9 (0.408) 11.4 (0.121) 13.9 (0.032) 21.2 (0.042)

GAC5% 323 (1.557) 250 (3.934) 85.5 (0.908) 27.4 (0.063) 78.1 (0.153)

GAC 4% 333 (1.609) 244 (3.838) 101 (1.073) 52.9 (0.122) 83.3 (0.164)

GAC 3% 385 (1.856) 213 (3.348) 92.6 (0.984) 59.2 (0.137) 49.8 (0.098)

GAC 2% 313 (1.508) 244 (3.838) 139 (1.476) 86.2 (0.199) 96.2 (0.189)

GAC 1% 370 (1.788) 227 (3.576) 196 (2.084) 101 (0.233) 145 (0.284)

GAC 0% 294 (1.420) 208 (3.278) 122 (1.296) 175 (0.405) 156 (0.307) CACa 25.1 (0.121) 20.9 (0.252) 219 (2.334) 178 (0.411) 129 (0.250)

Values in parentheses are presented in mmol/g. aThe experimental data were published in [16] with the same operation conditions. AR1, acid red 1; CAC, commercial activated carbon; GAC, glucose-activated carbon; GH, glucose-derived hydrochar; MG5, methylene green 5.

A 200 GH 5% A 180 180 GH 4% GH 5% 160 160 GH 3% GH 4% GH 2% GH 3% 140 140 GH 1% GH 2% GH 0% GH 0% 120 120 Langmuir fit Langmuir fit 100 100 (mg/g)

(mg/g ) 80 e

e 80 q q 60 60 40 40 20 20 0 0 0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000 1100 C (mg/l) e Ce (mg/l) B 200 B 180 GAC 5% 180 160 GAC 4% GAC 3% 160 140 GAC 2% 140 GAC 1% 120 GAC 0% 120 Langmuir fit 100 100 (mg/g)

e 80 q (mg/g )

e 80 q 60 60 GAC 5% GAC 4% 40 40 GAC 3% GAC 2% 20 GAC 1% 20 GAC 0% 0 Langmuir fit 0

0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000

Ce (mg/l) Ce (mg/l)

Figure 7: Adsorption isotherms of phenol by various (A) glucose Figure 8: Adsorption isotherms of methylene green 5 (MG5) by hydrochar and (B) activated carbon (AC) samples. various (A) glucose hydrochar and (B) activated carbon (AC) samples. H. N. Tran et al.: Activated carbon derived from spherical hydrochar functionalized with triethylenetetramine 573

A 160 valid for the GH0% sample, which has rich oxygen-bearing GH 5% groups on its surface and possesses the lowest adsorp- 140 GH 4% tion capacity (approximately 11 mg/g), and is consistent GH 3% 120 GH 2% with the findings of other researchers [31–33]. Further- GH 1% more, increasing TETA ratios reduces adsorption capaci- 100 GH 0% ties (Table 2 and Figure 7) because of an overabundance Langmuir fit 80 of amino groups linked to a smaller surface, which can (mg/g) e q π π 60 weaken the – interactions between the aromatic rings of the adsorbate and the adsorbent matrix [34]. 40 For hydrogen-bonding interactions (i.e. physical 20 adsorption), it is necessary to classify the types of H-bonding formation. First, hydrogen bonding occurs between 0 hydrogen from the hydroxyl group of phenol, and the 0 100 200 300 400 500 600 700 800 900 1000 oxygen complex on the surface of the adsorbent. Lorenc- C (mg/l) e Grabowska and colleagues [29] proposed that if hydrogen

160 bond formation is the mechanism of phenol adsorption, a B GAC 5% strong competition between water and phenol molecules 140 GAC 4% at low phenol concentration leads to reduced adsorption GAC 3% 120 GAC 2% capacity [30]; thus, water molecules are much more com- GAC 1% petitive in adsorbing on these groups compared with the 100 GAC 0% Langmuir fit hydrophobic phenol molecules [35]. However, concave 80 downward adsorption isotherms were obtained for all (mg/g) e q 60 the GH and GAC samples (Figure 7), indicating only a weak competition between phenol and water adsorption. 40 Therefore, the hydrogen bond interaction in this study 20 is negligible, evidenced in particular by the low adsorp-

tion capacity of GH0% (approximately 11 mg/g). Second, 0 hydrogen bonding occurs between the nitrogen groups 0 100 200 300 400 500 600 700 800 900 1000 on the adsorbent surface and the hydroxyl group of the C (mg/l) e phenol; similar results are available in the literature [34]. Additionally, amine groups have a dominant role in the Figure 9: Adsorption isotherms of acid red 1 (AR1) by various (A) glucose hydrochar and (B) activated carbon (AC) samples. adsorption of phenolic compounds onto amine-modified hyper-crosslinked polymeric resin through hydrogen bonding. This mechanism can explain why the adsorption not relevant to this study, and thus are not discussed. capacities of GH1%−2% and GAC1%−2% were higher than GH0%

This is because the amine groups on the surface of adsor- and GAC0% (Table 2 and Figure 7), respectively. + bents can be protonated to –NH3 under an acidic condi- The phenol adsorption capacity of GAC0% (approxi- tion (solution pH controlled at approximately 5.0), but the mately 122 mg/g) was extremely higher than that of GH0% phenol, a very weak acid, exists primarily in a nondissoci- because of the pore filling mechanism, indicating that ated form in water because its pKa is approximately 9.89. the micropore filling was the other central mechanism of

Dispersion interactions between the π-electron in phenol adsorption onto GAC0%. Phenol adsorption mainly phenol and the π-electron in carbon can be the dominant occurs in micropores smaller than 1.4 nm, and the average mechanism of adsorption in aromatic inorganic com- pore width of the GAC0% sample in this study was 2.58 nm pounds. However, the π–π interaction is greatly affected by (Table 1); an analogous result has been noted in prior the π-electron density in the basal plane of the carbonous studies [16, 29]. Furthermore, nitrogenation introduces materials. The abundance of oxygen- and nitrogen-con- nitrogen-containing functional groups onto the carbon taining functional groups (i.e. the electron-withdrawing surface, such as –NH2, pyridinic, pyrrolic, and quaternary groups) on the surface of carbonous materials can result nitrogen groups [14], which enhance the basicity of the in a drop in π-electron density and adsorption sites in carbon surface [36]. Phenols are more strongly removed the basal planes (graphene layers) because of π-electron from a solution by a basic carbon than an acidic carbon localization and withdrawal [31]. This explanation is [37]; therefore, the samples functionalized with TETA 574 H. N. Tran et al.: Activated carbon derived from spherical hydrochar functionalized with triethylenetetramine

always exhibited higher adsorption capacities than did AC (129 mg/g), indicating that GAC1% became an excel- those without TETA (Table 2 and Figure 7). lent adsorbent.

The adsorption capacity of MG5 onto GAC0% (approximately 157 mg/g) was higher than that of GAC1%−5% (approximately 27.4–101 mg/g), which might be attributed 4 Conclusions to the higher cation exchange of GAC0% than GAC1%−5%; this coincides with the higher Na content on the GAC0% (5.61%) Synthesizing hydrochar and AC with a TETA-glucose than GAC1% (3.80%) surfaces. As Romero and others [19] mixture substantially changes the morphology of the reported, ion exchange plays a critical role in the adsorp- modified hydrochar samples. Specifically, increasing tion of MG1 onto NaOH-AC spheres with abundant surface- TETA concentrations leads to a diminished surface area grafted functional groups. (SBET), non-micropore volume, and total pore volume in

The adsorption capacity of GAC0% (175 mg/g) for glucose ACs, and an increased micropore volume. MG5 was nearly the same as that of commercial AC The adsorption capacities of the GH samples toward (178 mg/g) under the same experimental conditions; as inorganic and organic compounds were considerably Figure S5 illustrates, this study similarly found an excel- enhanced with TETA modification. Specifically, increas- lent relationship between the maximum adsorption capac- ing TETA concentrations resulted in decreased adsorption o ity (Q max) of MG5 by GACs and non-micropore surface area capacities for the compound contaminants, with the GH1% (R2 = 0.97) and nonmicropore volume (R2 = 0.96). These sample exhibiting the highest affinity toward the contami- o correlated values demonstrate that the adsorption of MG5 nants. The maximum adsorption capacities (Q max) of GH1% by GACs was governed by non-micropore filling. were ordered as follows: phenol (1.456 mmol/g) > Cu2 + The maximum adsorption capacities of the TETA- (1.334 mmol/g) > Pb2 + (0.619 mmol/g) > MG1 modified hydrochar samples were nearly equal to those (0.156 mmol/g) > AR1 (0.071 mmol/g). of the AC samples with TETA: GH5% (27.5 mg/g) ≈ GAC5% The GAC samples exhibited even higher adsorption

(27.4 mg/g), GH4% (55.9 mg/g) ≈ GAC4% (52.9 mg/g), and capacities than did their GH precursors and the com-

GH3% (53.5 mg/g) ≈ GAC2% (59.2 mg/g). This result indicates mercial AC, with GAC1% reaching the highest adsorption o that TETA has a positive effect on the adsorption enhance- capacity for most of the contaminants. The Q max values ment of MG5. of GAC1% for the contaminants were ordered as follows: Clearly, the affinities of the GH samples toward Cu2 + (3.576 mmol/g) > phenol (2.084 mmol/g) > Pb2 + AR1 were lower than those toward phenol and MG5, (1.788 mmol/g) > AR1 (0.284 mmol/g) > MG1 because the negative charge for AR1 can generate the (0.233 mmol/g). These values are notably higher than the repulsion forces of the functional groups (i.e. –COO −) corresponding values of commercial AC. on the hydrochar surface. If the amine groups (–NH2) on the hydrochar samples with TETA are completely proto- Acknowledgments: This work was financially supported + nated to –NH3 under an acidic condition (pH = approxi- by Chung Yuan Christian University (CYCU), Taiwan. mately 5.0), then the adsorption capacity of the The first author would like to thank CYCU for the Distin- hydrochar samples with TETA are enhanced because guished International Graduate Students (DIGS) scholar- of electrostatic attraction. In this study, however, the ship to pursue his doctoral studies. amine groups (–NH2) on the TETA-modified hydrochar + samples were not protonated to –NH3 in the solution; Conflict of interest statement: The authors declare that no therefore, the adsorption mechanisms of AR1 onto the competing financial or other conflicts exist. hydrochar sample with TETA did not involve the amine groups through an anion exchange process, and the hydrogen bonding interactions between the nitrogen- References and oxygen-containing functional groups on the dye molecule and the adsorbents’ surface might be respon- [1] Freedonia, World activated carbon - industry market research, sible for the adsorption mechanisms [38, 39]. Similar to market share, market size, sales, demand forecast, market leaders, company profiles, industry trends. Industry Studies and the adsorption of MG5, GAC0% (approximately 156 mg/g) Freedonia Focus Report, 2014. also exhibited a higher affinity toward AR1 than [2] Romero-Anaya AJ, Ouzzine M, Lillo-Ródenas MA, Linares-Solano GAC1%−5% (approximately 49–145 mg/g); moreover, the A. Carbon 2014, 68, 296–307. adsorption of AR1 by the GAC0% (156 mg/g) and GAC1% [3] Jain A, Balasubramanian R, Srinivasan MP. Chem. Eng. J. 2016, (145 mg/g) samples was higher than that of commercial 283, 789–805. H. N. Tran et al.: Activated carbon derived from spherical hydrochar functionalized with triethylenetetramine 575

[4] Titirici MM, White RJ, Falco C, Sevilla M. Energy. Environ. Sci. 2012, 5, 6796–6822. Bionotes [5] Bedin KC, Martins AC, Cazetta AL, Pezoti O, Almeida VC. Chem. Eng. J. 2016, 286, 476–484. Hai Nguyen Tran [6] Mwenya T, Fan H, Dai H, Li M. Int. J. Photoenergy 2016, 2016, 1–5. [7] Demir-Cakan R, Baccile N, Antonietti M, Titirici M-M. Chem. Mater. 2009, 21, 484–490. [8] Wang X, Liu J, Xu W. Colloids Surf. A 2012, 415, 288–294. [9] Chizari K, Vena A, Laurentius L, Sundararaj U. Carbon 2014, 68, 369–379. Hai Nguyen Tran is currently a PhD student at the Environmental [10] Karnitz Jr O, Gurgel LV, de Melo JC, Botaro VR, Melo TM, de Engineering Department, Chung Yuan Christian University, Taiwan. Freitas Gil RP, Gil LF. Bioresour. Technol. 2007, 98, 1291–1297. His recent publications focused on (1) the removal of cadmium from [11] Niasar HS, Li H, Kasanneni TVR, Ray MB, Xu C. Chem. Eng. J. aqueous solutions using orange peel-derived biochar produced 2016, 293, 189–199. from different pyrolysis temperatures and times, (2) comparison [12] Gurgel LVA, Gil LF. Water Res. 2009, 43, 4479–4488. of the calculation results of the thermodynamic parameters in [13] Gurgel LVA, Gil LF. Carbohydr. Polym. 2009, 77, 142–149. the adsorption study by different methods, (3) preparation and [14] Barsanescu A, Buhaceanu R, Dulman V, Bunia I, Neagu V. characterization of various adsorbents (i.e. biosorbent, hydrochar, J. Appl. Polym. Sci. 2004, 93, 803–808. biochar, spherical carbon, activated carbon, surfactant modified [15] Sing KS. Pure Appl. Chem. 1985, 57, 603–619. zeolite, and Mg-Al LDHs intercalated with phenylalanine, tyrosine, [16] Huang FC, Lee C-K, Han Y-L, Chao W-C, Chao H-P. J. Taiwan Inst. and serine) for the removal of organic and inorganic contaminants Chem. Eng. 2014, 45, 2805–2812. from aqueous solutions, (4) preparation of activated carbon from a [17] Sevilla M, Fuertes AB. Chem. Eur. J. 2009, 15, 4195–4203. new chemical activation method. https://www.researchgate.net/ [18] Sevilla M, Fuertes AB. Carbon 2009, 47, 2281–2289. profile/Hai_Tran16. [19] Romero VJM, Tsai JF, Doma Jr, BT, Chao HP. Sustain. Environ. Res. 2014, 24, 339–347. [20] Song X, Gunawan P, Jiang R, Leong SS, Wang K, Xu R. J. Hazard. Fu-Chuang Huang Mater. 2011, 194, 162–168. [21] Lai L, Huang G, Wang X, Weng J. Carbon 2010, 48, 3145–3156. [22] Salehi E, Madaeni SS, Rajabi L, Vatanpour V, Derakhshan AA, Zinadini S, Ghorabi S, Ahmadi Monfared H. Sep. Purif. Technol. 2012, 89, 309–319. [23] Zhu J, Yang J, Deng B. J. Hazard. Mater. 2009, 166, 866–872. [24] Brady D, Duncan JR. Enzyme Microb. Technol. 1994, 16, 633–638. [25] Tran HN, You SJ, Chao HP. Waste Manage. Res. 2016, 34, Fu-Chuang Huang completed his PhD in Environmental Engineering 129–138. at National Central University, Chung-Li, Taiwan. Currently, he is a [26] Chand P, Shil AK, Sharma M, Pakade YB. Int. Biodeterior. Biode- faculty member at Nanya Institute of Technology, Chung-Li, Taiwan. grad. 2014, 90, 8–16. His research interests are in soil pollution remediation, adsorption, [27] Xiao B, Thomas KM. Langmuir 2005, 21, 3892–3902. wastewater treatment, and environmental education. [28] Petrov N, Budinova T, Khavesov I. Carbon 1992, 30, 135–139. [29] Lorenc-Grabowska E, Gryglewicz G, Diez MA. Fuel 2013, 114, 235–243. Chung-Kung Lee [30] Moreno-Castilla C. Carbon 2004, 42, 83–94. [31] Radovic LR, Baker FS. Carbon 1997, 35, 1339–1348. [32] Mahajan OP, Moreno-castilla C, Walker PL. Sep. Sci. Technol. 1980, 15, 1733–1752. [33] Salame II, Bandosz TJ. J. Colloid Interface Sci. 2003, 264, 307–312. [34] Li A, Zhang Q, Wu H, Zhai Z, Liu F, Fei Z, Long C, Zhu Z, Chen J. Adsorp. Sci. Technol. 2004, 22, 807–819. [35] Franz M, Arafat HA, Pinto NG. Carbon 2000, 38, 1807–1819. [36] Yang G, Chen H, Qin H, Feng Y. Appl. Surf. Sci. 2014, 293, Chung-Kung Lee completed his PhD in Chemical Engineering at 299–305. National Central University, Chung-Li, Taiwan. He has published [37] Terzyk AP. J. Colloid Interface Sci. 2003, 268, 301–329. more than 65 research articles as first and co-author in SCI journals. [38] Li M, Wang H, Wu S, Li F, Zhi P. RSC Adv 2012, 2, 900–907. He received his BSc in Chemical Engineering from Taipei Institute [39] Ahmad R, Kumar R. J. Chem. Eng. Data 2010, 55, 3489–3493. of Technology, Taipei, Taiwan and his MSc in Chemical Engineering from National Central University, Chung-Li, Taiwan. Currently, he is a faculty member at University of Vanung, Chung-Li, Taiwan. His Supplemental Material: The online version of this article research interests are in fractal and chaos analysis, adsorption, (DOI: 10.1515/gps-2016-0178) offers supplementary material, catalysis and wastewater treatment. https://www.researchgate.net/ available to authorized users. profile/Chung_Kung_Lee. 576 H. N. Tran et al.: Activated carbon derived from spherical hydrochar functionalized with triethylenetetramine

Huan-Ping Chao

Huan-Ping Chao completed his PhD in Graduate Institute of Environmental Engineering, National Central University, Chung-Li, Taiwan. He has published more than 35 research articles as first and co-author in SCI journals. Currently, he is a faculty member at Department of Environmental Engineering, Chung Yuan Christian University, Chung-Li, Taiwan. His research interests are in volatiliza- tion of organic compounds from water, and development of new adsorbents to remove contaminants in wastewater. https://www. researchgate.net/profile/Huan-Ping_Chao.